Recherche avancée

Sur d’autres sites (5384)

• Revision 30177 : un plugin opensearch pour proposer d’ajouter son site aux moteurs ...

  24 juillet 2009, par brunobergot@… — Log

  un plugin opensearch pour proposer d’ajouter son site aux moteurs personnalisés des navigateurs qui supportent ce protocole

• Using FFMPEG, how do I convert and publish video to main site in right time ?

  30 septembre 2018, par Izak

  I am building an online video platform. Using FFMPEG, I am able to upload a MP4 video to the server, convert it and play it. The problem is that while FFMPEG is still converting the video and has not chosen a thumbnail in the backend, the video is already published to the main site (frontend)...users see an ugly video placeholder and cannot play the video.

  I think I need to add a timer between conversion and publishing of the video to frontend ? Or how do you think I should solve this problem ?

  Thanks for your insight...(any shared code is also welcome)

  Isaac

• ffmpeg download mp3 in chunks slower than whole file

  6 août 2021, par loretoparisi

  I'm programmatically downloading in Python mp3 file's as wav chunks, seeking at position with ss and t with ffmpeg resulting in a sequence of ffmpeg commands

  


  ffmpeg -i https://file-examples-com.github.io/uploads/2017/11/file_example_MP3_5MG.mp3 -acodec pcm_s16le -ac 1 -ar 44100 -ss 0:08:55.780000 -t 0:01:00.000000 -sn -vn -y -f wav pipe:1
ffmpeg -i https://file-examples-com.github.io/uploads/2017/11/file_example_MP3_5MG.mp3 -acodec pcm_s16le -ac 1 -ar 44100 -ss -ss 0:09:55.780000 -t 0:01:00.000000 -sn -vn -y -f wav pipe:1
ffmpeg -i https://file-examples-com.github.io/uploads/2017/11/file_example_MP3_5MG.mp3 -acodec pcm_s16le -ac 1 -ar 44100 -ss -ss -ss 0:10:55.780000 -t 0:01:00.000000 -sn -vn -y -f wav pipe:1
...


  


  executed via concurrent.futures.ThreadPoolExecutor in this way :

  


  with concurrent.futures.ThreadPoolExecutor(max_workers=NTHREADS) as executor:
        for i,cmd in enumerate(process_list):
            futures.append(executor.submit(downloader.chunk_download,(cmd,i)))

        for future in concurrent.futures.as_completed(futures):
            try:
                completed.append(future.result())
            except Exception as e:print(e)
    completed.sort() 
    for k,c in enumerate(completed):
        if not k:
            waveform = c[1]
        else:
            waveform = np.append(waveform,c[1])


  


  where chunk_download is concat the chunks into the whole waveform

  


  def chunk_download(self,args):
        cmd=args[0]
        i=args[1]
        print('Downloading chunk n° %i' % i)

        print( ' '.join(cmd))
        process = subprocess.run(cmd,
                            stdout=subprocess.PIPE,
                            stderr=subprocess.PIPE,
                            bufsize=10**8)
        #buffer, stderr = process.communicate()
        buffer = process.stdout
        stderr = process.stderr
        print('chunk n° %i DOWNLOADED' % i)
        # convert to signal
        chunk_waveform = np.frombuffer(buffer=buffer, dtype=np.uint16, offset=8*44)
        chunk_waveform = chunk_waveform.astype(self.dtype)  
        print(len(chunk_waveform)/44100) 

        return i,chunk_waveform


  


  this works ok, but if I download the whole mp3, executing the command in cmd like :

  


  ffmpeg -i https://file-examples-com.github.io/uploads/2017/11/file_example_MP3_5MG.mp3 -acodec pcm_s16le -ac 1 -ar 44100 -sn -vn -y -f wav pipe:1


  


  It will take less time than the concurrent chunks download for the same file length, while I would expect the opposite. Could be this related to ffmpeg threading option (param --thread ?). If not, why the concurrent chunks download operation is slower ?